Effect of nasendoscopy procedure on acoustic features of voice and speech

Abstract

Nasendoscopy procedure enables direct visualization of voice and speech anatomy. The procedure passes a flexible scope transnasally to observe velopharyngeal mechanism or the larynx during speech and phonation. This imaging technique is commonly used simultaneously with audio recordings. Studies have identified important statistically-significant relationships between imaging-based and acoustic measures related to voice and speech quality. However, the effect of the nasendoscopy procedure itself on specific acoustic features is rarely being considered. Nasendoscopy is likely to cause some speakers to deviate from their typical vocal function by inducing stress and tension during recordings. Furthermore, the flexible nasendoscope can also change the sound and airflow behavior inside the vocal tract. The current study attempts to quantify how nasendoscopy influences acoustic features using a physical model based on a child's vocal tractduring a sibilant sound. An electrolarynx was used to provide a constant sound source with the input airflow to the model. Audio recordings were captured while retracting the nasendoscope from the level of the hypopharynx. Results show that compared to the same data taken without nasendoscopy, the scope can change measures of signal-to-noise and cepstral-peak-prominence by 15% and 7%, respectively. The nasendoscope had a minimal effect (<3%) on the SPL measure.